LiteLLM validates provider-specific API key env vars before making calls.
When using custom headers, auto-set ANTHROPIC_API_KEY or OPENAI_API_KEY
to pass validation (actual auth still uses extra_headers).
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add LLM_API_KEY_HEADER for providers using non-standard auth headers
- Support Xiaomi MiMo (uses "api-key" header instead of "x-api-key")
- Add extra_headers to LiteLLM calls when custom header is configured
- Document LITELLM_ANTHROPIC_DISABLE_URL_SUFFIX for full path URLs
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Support both OpenAI-compatible and Anthropic-compatible endpoints.
User must specify full model name with provider prefix.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
LiteLLM requires model names in format "provider/model-name".
When LLM_API_BASE is set, automatically prefix model with "openai/"
if no provider prefix is present.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add com.unity.inputsystem package dependency
- Ignore TextMesh Pro auto-generated font assets in gitignore
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Replace openai with litellm for unified LLM interface
- Support 100+ providers: OpenAI, Anthropic, Gemini, Azure, Ollama, etc.
- Add custom API base URL support (LLM_API_BASE)
- Add .env file support with python-dotenv
- Add .env.example configuration template
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Unity client features:
- WebSocket connection via NativeWebSocket
- 2.5D agent visuals with programmatic placeholder sprites
- Billboard system for sprites and UI elements
- Floating UI panels (name, HP, energy bars)
- Speech bubble system with pop-in animation
- RTS-style camera controller (WASD + scroll zoom)
- Editor tools for prefab creation and scene setup
Scripts:
- NetworkManager: WebSocket singleton
- GameManager: Agent spawning and event handling
- AgentVisual: 2.5D sprite and UI creation
- Billboard: Camera-facing behavior
- SpeechBubble: Animated dialogue display
- CameraController: RTS camera with UI input detection
- UIManager: HUD and command input
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add OpenAI-compatible LLM integration for agent dialogue
- Enhance survival mechanics with energy decay and feeding system
- Update frontend debug client with improved UI
- Add .gitignore rules for Unity and Serena
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add Player and Boss data models (models.py)
- Implement command processing via regex matching (attack/heal/status)
- Add boss counter-attack mechanism (15 dmg per player attack)
- Add player death/respawn system (lose half gold, respawn full HP)
- Update frontend with boss HP bar, player stats panel, quick action buttons
- Add colored event log (red for attack, green for heal)
- Change port from 8000 to 8080
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add FastAPI backend with WebSocket support
- Implement ConnectionManager for client connections
- Create GameEngine with async game loop (2s tick)
- Add RuleBasedAgent for keyword-based responses
- Define Pydantic schemas for GameEvent protocol
- Create debug frontend dashboard for testing
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
